Small Molecule Reversible Inhibitors of Bruton's Tyrosine Kinase (BTK

Article pubs.acs.org/jmc

Small Molecule Reversible Inhibitors of Bruton’s Tyrosine Kinase (BTK): Structure−Activity Relationships Leading to the Identification of 7‑(2-Hydroxypropan-2-yl)-4-[2-methyl-3-(4-oxo-3,4dihydroquinazolin-3-yl)phenyl]‑9H‑carbazole-1-carboxamide (BMS935177) George V. De Lucca,*,† Qing Shi,† Qingjie Liu,† Douglas G. Batt,† Myra Beaudoin Bertrand,† Rick Rampulla,† Arvind Mathur,† Lorell Discenza,∥ Celia D’Arienzo,∥ Jun Dai,∥ Mary Obermeier,∥ Rodney Vickery,∥ Yingru Zhang,∥ Zheng Yang,∥ Punit Marathe,∥ Andrew J. Tebben,§ Jodi K. Muckelbauer,§ ChiehYing J. Chang,§ Huiping Zhang,† Kathleen Gillooly,‡ Tracy Taylor,‡ Mark A. Pattoli,‡ Stacey Skala,‡ Daniel W. Kukral,⊥ Kim W. McIntyre,‡ Luisa Salter-Cid,‡ Aberra Fura,∥ James R. Burke,‡ Joel C. Barrish,† Percy H. Carter,† and Joseph A. Tino*,† †

Immunosciences Discovery Chemistry, ‡Immunoscience Discovery Biology, §Molecular Structure and Design, Molecular Discovery Technologies, ∥Metabolism and Pharmacokinetic Department, Pharmaceutical Candidate Optimization, and ⊥ECTR/CTTO Imaging Department, Bristol-Myers Squibb Research and Development, P.O. Box 4000, Princeton, New Jersey 08543, United States S Supporting Information *

ABSTRACT: Bruton’s tyrosine kinase (BTK) belongs to the TEC family of nonreceptor tyrosine kinases and plays a critical role in multiple cell types responsible for numerous autoimmune diseases. This article will detail the structure−activity relationships (SARs) leading to a novel second generation series of potent and selective reversible carbazole inhibitors of BTK. With an excellent pharmacokinetic profile as well as demonstrated in vivo activity and an acceptable safety profile, 7-(2-hydroxypropan-2-yl)-4-[2-methyl-3-(4-oxo-3,4-dihydroquinazolin-3-yl)phenyl]-9H-carbazole-1-carboxamide 6 (BMS935177) was selected to advance into clinical development.

■

INTRODUCTION Kinases continue to be a very active area for drug discovery, targeting treatments in a wide range of therapeutic areas.1 Bruton’s tyrosine kinase (BTK) is a nonreceptor tyrosine kinase expressed in all hematopoietic cells except plasma and T cells. Although only one of the five mammalian TEC family members, BTK is the most important from a target perspective, since it is critical for B cell signaling through the B cell receptor (BCR).2 In addition to B cells, BTK is vital for signaling in low affinity activating Fcγ receptors (e.g., FcγRIII and FcγRIIa) in monocytic cells. BTK also regulates the expression of proinflammatory cytokines, chemokines, and cell adhesion molecules in response to receptor activation through FcεRI signaling in mast cells and basophils.3 RANK signaling, which controls osteoclastogenesis from monocytic precursors, is also BTK-dependent.4 B cells are essential to the pathogenesis of autoimmune diseases such as rheumatoid arthritis (RA) as demonstrated by the clinical results for the B cell-depleting anti-CD20 antibody rituximab.5 In addition to producing autoantibodies, B cells can control autoimmunity as sources of proinflammatory chemo© 2016 American Chemical Society

kines and cytokines, such as IL-6 and TNFα, and through their role as antigen-presenting cells.6 Because BTK kinase activity is necessary for BCR-dependent proliferation of B cells as well as production of proinflammatory cytokines and costimulatory molecules (e.g., CD86), pharmacological inhibition of BTK is expected to affect several pathways by which B cells may mediate autoimmunity but without depleting B cells. Pharmacological and genetic studies in animal models strengthen the case that BTK inhibitors would be efficacious against autoimmune indications such as RA or systemic lupus erythematosus (SLE). BTK-deficient mice are less sensitive to both the (NZB X NZW)F1 model of lupus and the collageninduced arthritis (CIA) model.7 BTK inhibitors reduce disease progression in the collagen antibody-induced arthritis (CAIA) and CIA models, in addition to murine lupus models.8 The vital role of BTK in B cells is also observed in human x-linked agammaglobulinemia (XLA) immunodeficiency. XLA patients Received: May 10, 2016 Published: August 17, 2016 7915

DOI: 10.1021/acs.jmedchem.6b00722 J. Med. Chem. 2016, 59, 7915−7935

Journal of Medicinal Chemistry

Article

with improved oral exposure, kinase selectivity, and BTK potency, leading to the discovery of the clinical development candidate 6 (BMS-935177).17b

have inactivating mutations in the BTK gene leading to a loss of mature B cells and circulating antibodies.9 A key structural feature of the BTK kinase domain is the noncatalytic cysteine (Cys481) residue in the extended hinge region.10 Numerous companies have targeted this residue to prepare irreversible BTK inhibitors, with the most successful being ibrutinib (1; Pharmacyclics/J&J), which is approved for various non-Hodgkin’s lymphomas (Figure 1).11,12 Several

■

RESULTS AND DISCUSSION As we have reported previously, X-ray structural data for a member of the carbazole carboxamide series showed that the carbazole NH and both the carbonyl and NHs of the primary amide formed critical hydrogen bonds to the key hinge region of BTK.17a It was also noted that a small ortho-substituent on the C-4 phenyl ring improved both BTK potency and kinase selectivity when compared to the parent unsubstituted analog. In addition, X-ray structures showed the C-7 substituent bound in the solvent exposed extended hinge region of BTK.17a Several initial issues for the carbazole series required optimization, including kinase potency and selectivity, as well as oral exposure in mouse. Our efforts to address these concerns involved the use of a combination of focused library and individual analog synthesis as outlined in Figure 3. The 4fluorophenyl amide of 5 was replaced with a variety of alkyl, aryl, and heteroaryl amides as well as ureas and carbamates using the aniline 7a. To explore the SAR at the C-7 position, the piperazine of amide of 5 was replaced with a variety of alkyl, aryl, and heteroaryl amides starting from the acid 8a. The acid substituent of a 8a precursor was converted to aniline 9a (see Experimental Section) which was used as the starting material to further explore the SAR at C-7 to give 9b, incorporating a variety of alkyl anilines as well as a range of alkyl, aryl, and heteroaryl amides, ureas, and carbamates. From this thorough exercise two analogs were noteworthy, 10 and 11 (Table 1). The polarity of the piperazine amide C-7 substituent coupled with the large number of hydrogen bond donors and acceptors in 5 led to low permeability as measured in PAMPA, contributing to low oral exposure in mouse (Table 1). The ortho-pyridyl analog 10, designed to form an intramolecular Hbond with the amide NH to improve permeability, maintained good BTK potency with improved PAMPA and oral exposure as shown in Table 1, although selectivity over the SRC family of kinases (LCK as an example found on T-cells) was still unacceptable. The goal of >50-fold selectivity over the SRC family was set to minimize the chances of being too immunosuppressive in vivo by targeting both B- and T-cells. With recognition that the C-7 substituents could potentially bind in the solvent exposed extended hinge region of BTK, modification of the side chain could be used to “tune” the properties of inhibitors. Analog exploration at C-7 (Figure 3) often showed a dramatic effect on kinase selectivity and potency. However, many of these analogs were of high molecular weight with low oral plasma concentrations in mouse PK studies (results not shown). The amino acid at the

Figure 1. BTK irreversible inhibitors.

additional BTK inhibitors have entered clinical trials for the treatment of autoimmune indications, including covalent inhibitors from Celgene (2), Hanmi (3), and Principia (4, structure unknown, 2014 press release),13 with the most advanced appearing to be 2, currently in phase II trials in RA (Figure 1).14 Efficacy results for these and other compounds entering clinical trials are eagerly awaited, since small molecule inhibitors of this mechanism have the potential to compete with biologics for the treatment of RA and several other autoimmune diseases. In addition to these covalent inhibitors, recent reports from Roche and Merck show continued interest in reversible BTK inhibitors.15,16 We have previously disclosed the discovery of a novel carbazole series of BTK inhibitors, as well as the initial structure−activity relationships (SARs) leading to 5 (Figure 2).17 Compound 5 inhibited BTK with modest kinase selectivity. Having identified a novel carbazole series of BTK inhibitors, we began an effort to address the liabilities of 5 and similar analogs. In the current paper, we disclose carbazoles

Figure 2. Initial lead 5 and optimized compound 6. 7916



Article

Figure 3. Initial optimization of carbazole core series.

Table 1. In Vitro Potency, Selectivity, and Mouse PK of Carbazole Analogues

a

IC50 values are shown as mean values of three determinations. bFold enzyme IC50 selectivity. cPAMPA values were determined at pH 7. dCMAX (maximum concentration) after 10 mg/kg oral dosing in male Balb/C mice with PEG300/TPGS/ethanol (90/5/5 v/v/v) vehicle. eAUC0−24h (area under the curve) after 10 mg/kg oral dosing in male Balb/C mice with PEG300/TPGS/ethanol (90/5/5 v/v/v) vehicle. fNull = compound was not detected.

C-terminal end of the hinge in BTK, Ala478, is often an acidic residue in SRC kinase family members such as LCK (Glu320). We predicted that a less basic C-7 group could improve selectivity over the piperazine amide. In order to keep the

molecular weight low, compound 11 was prepared and found to be a potent BTK inhibitor with improved oral plasma exposure and LCK kinase selectivity (Table 1). However, analogs with a pendent meta-aryl or heteroaryl amide 7917



Article


a

IC50 values are shown as mean values of at least three determinations. bFold enzyme IC50 selectivity. cCMAX (maximum concentration) after 10 mg/kg oral dosing in male Balb/C mice with PEG300/TPGS/ethanol (90/5/5 v/v/v) vehicle. dAUC0−24h (area under the curve) after 10 mg/kg oral dosing in male Balb/C mice with PEG300/TPGS/ethanol (90/5/5 v/v/v) vehicle. eIC50 values are shown as single determinations. fND = not determined. 7918



Article


compd 22 23

R

BTKa IC50 (nM)

H F

e

44 4±2

LCK/BTKb

Ramos,a IC50 (nM) e

32× 210×

410 30 ± 2

CMaxc (nM) f

ND 2100

AUCd (μM·h) NDf 19

a

IC50 values are shown as mean values of at least three determinations. bFold enzyme IC50 selectivity. cCMAX (maximum concentration) after 10 mg/kg oral dosing in male Balb/C mice with PEG300/TPGS/ethanol (90/5/5 v/v/v) vehicle. dAUC0−24h (area under the curve) after 10 mg/kg oral dosing in male Balb/C mice with PEG300/TPGS/ethanol (90/5/5 v/v/v) vehicle. eIC50 values are shown as single determinations. fND = not determined.

substituent on the C-4 phenyl, as in 10 and 11, showed hydrolysis to an aniline metabolite as the major metabolic pathway in human and rat liver microsomes. In addition, in the case of 11, concomitant N-dealkylation of the isopropyl group led to the corresponding bis-aniline metabolite, observed in liver microsomes and as a circulating metabolite after oral dosing in mice. Upon multiple day dosing in mice, compound 11 showed toxicity in vivo that could be attributed to the bisaniline metabolite (data not shown). Predicting the extent of amide hydrolysis in humans based on preclinical data is potentially challenging,18 supporting the decision to stop further progression of these analogs. To avoid any possible aniline release, numerous medicinal chemistry approaches to stabilize amide hydrolysis were investigated. The most direct approach, synthesis of the cyclic amides, was also the most effective. The five-membered ring lactam 12 maintained BTK potency with acceptable permeability (PAMPA Pc 494 nm/s, Table 2). As expected, 12 was stable to hydrolysis in vivo; however the plasma concentration in mouse was similar to 5, which also contained the piperazine amide. It should be noted that the mouse liver microsome metabolic stability25e of the three compound 5, 10, and 12 containing the piperazine amide was comparable. It was surprising that the exposure for lactam 12 was significantly worse than that of ortho-pyridyl analog 10, designed to form an intramolecular H-bond with the amide NH, which points to the difficulty of predicting exposure based solely on one parameter, such as numbers of hydrogen bonds. The SAR results from our previous C-7 analogs prepared for compound 5 (outlined in Figure 3) suggested that small polar nonbasic or weakly basic groups would be best for potency, selectivity, and PAMPA. Combining the C-4 lactam with a tertiary carbinol as a C-7 substituent led to analog 13 which maintained BTK potency and improved potency in the Ramos B cell calcium flux assay (Table 2). In order to further optimize 13, a second survey of alternative C-7 substituents was undertaken focusing on small polar groups. Ultimately, no other substituent was found to give better overall properties than the carbinol. An investigation of substituents on the phenyl ring of the lactam identified the fluoro-substituted analog 14, which had improved BTK cell potency and mouse liver microsome metabolic stability compared to 13. Compound 14 showed significantly improved oral plasma concentrations in mouse compared to 12 (Table 2), suggesting that it was suitable for further in vivo studies.

BTK regulates antigen receptor (BCR) signaling in B cells, and one of the hallmarks of xid (x-linked immune defect) mice deficient in BTK is a defective neoantigen-induced antibody response. As a result, it was anticipated that activity against a keyhole limpet hemocyanin (KLH) induced antibody response in mice would provide an important pharmacodynamics (PD) readout of BTK inhibition.19 In this model, anti-KLH antibodies of the IgM isotype appear in the first week after challenge, followed by isotype switching to IgG anti-KLH antibodies evident by day 14. However, 14 showed only a nonstatistically significant decrease in IgG when dosed orally at 30 mg/kg b.i.d. in the chronic KLH mouse model. The weak suppression of the anti-KLH response, coupled with a nonoptimal PK profile, prohibited further advancement of compound 14. An extensive exploration of heterocyclic analogs of 14 was conducted to find compounds with improved potency and oral plasma concentrations (Table 2). BTK potency was maintained in several 6,6 bicyclic ring systems, such as in analogs 15 and 16, but they were either less LCK (or other kinase) selective or less potent in cells than 14. The unsubstituted quinazolinone 6 showed improved Ramos B cell potency with similar LCK selectivity to 14. Quinazolinones were substituted with an array of electron donating and withdrawing groups, as well as a range of both polar and lipophilic moieties. However, in none of these cases was the overall profile improved over the unsubstituted parent analog 6. For example, a fluoro scan on the quinazolinone ring provided 17−20. Although all of the fluoro analogs were potent BTK inhibitors, none showed significant improvements over 6. All five analogs were investigated in mouse PK studies at 10 mpk, showing improved oral plasma concentrations compared to 14 (Table 2). Quinazolinone analogs 21 (Table 2) and 22 (Table 3) were also prepared to investigate the requirements of the tertiary carbinol and the ortho-methyl substituent in the C-4 aryl linker, respectively. Compound 21 maintained BTK enzyme potency but showed lower oral plasma concentrations in mouse compared to 6. Analog 22 had a significant drop in BTK enzyme potency, highlighting the importance of the C-4 aryl ortho-methyl substituent. Attempts to obtain an X-ray cocrystal structure of 6 bound to BTK were unsuccessful. However, close analog 23 (Table 3) provided a 1.9 Å X-ray crystal structure bound to the active site of BTK (PDB code 5JRS), as shown in Figure 4. The carbazole NH and primary amide were bound to the hinge as expected,17a 7919



Article

further animal studies. Therefore, compound 6 was selected for detailed in vitro and in vivo characterization. A partial list of kinase selectivity data for compound 6 is summarized in Table 4. Compound 6 is a potent, reversible Table 4. Partial List of in Vitro Kinase Selectivity Data for Compound 6 kinase

IC50, μM

fold BTK selectivity

TEC BMX ITK TXK BLK TRKA HER4 TRKB RET LCK LYN SRC JAK2

0.013 0.024 0.096 0.180 0.020 0.030 0.050 0.10 0.11 0.15 0.16 3.19 0.37

5 9 28 67 7 11 18 35 40 55 60 1140 131

inhibitor of BTK (IC50 = 2.8 nM) and demonstrated good kinase selectivity when tested against a screening panel of 384 kinases at DiscoveRx (Fremont, CA; formally Ambit Biosciences),20 with only 16 kinases showing less than 15% control binding at 1000 nM. Consistent with this measure of selectivity, biochemical kinase assays against 60 kinases showed 6 to be more potent against BTK than any other kinase, including the other Tec family kinases (TEC, BMX, ITK, and TXK) over which the compound is between 5- and 67-fold selective. With the exception of 7-fold selectivity over BLK, expressed specifically in B cells, the compound showed greater than 50fold selectivity over the SRC family of kinases, including 1100fold selectivity over SRC itself. Other kinases inhibited with a potency less than 150 nM (50-fold selectivity) included TRKA, HER4, TRKB, and RET. Cell potency data for compound 6 are summarized in Table 5. In B cells stimulated through the BCR, 6 selectively inhibited several different readouts. Compound 6 inhibited calcium flux in human Ramos B cells (IC50 = 27 nM) and inhibited CD69 surface expression in peripheral B cells stimulated with antiIgM and anti-IgG. However, 6 had no effect on CD69 surface expression in B cells stimulated through the CD40 receptor with CD40 ligand. Against IgG-containing immune complexdriven low affinity activating Fcγ receptor (FcγRIIa and FcγRIII) end points in peripheral blood mononuclear cells (PBMCs), 6 effectively inhibited TNFα production with an IC50 value of 14 nM. To determine the whole blood potency, the ability of the compound to inhibit BCR-stimulated expression of CD69 on the surface of B cells in whole blood was measured in a FACS-based assay which allowed for gating on the B cells. Against this BTK-dependent end point, 6 showed mean IC50 values of 550 ± 100 (n = 11) and 2060 ± 240 nM (n = 3) in human and mouse whole blood, respectively. The somewhat more potent inhibition in human compared to mouse whole blood may reflect both an intrinsic human potency advantage and protein binding differences between species (Table 6). Some of the detailed in vitro profiling data for compound 6 are summarized in Table 6. Plasma protein binding for 6 was high for all species, with less than 1% free for human. The high

Figure 4. (A, B) Compound 23 X-ray crystal structure bound to BTK kinase domain: (A) electron density mapping of 23; (B) binding of 23 (PDB code 5JRS), including waters.

forming crucial hydrogen bonds with the backbone carbonyl of Met-477 and both the backbone NH of Met-477 and the carbonyl of Glu-475, respectively. In addition, the primary amide formed a third hydrogen bond to a water in the gate keeper region, which was linked to an extensive water network that filled the pocket. The fluorophenyl ring was nearly orthogonal to the carbazole, with the ortho fluoro filling a small hydrophobic pocket formed by Leu-528 at the base of the ATP pocket and the backbone of Cys-481. The quinazolinone ring was in turn orthogonal to the C-4 phenyl linker, with the carbonyl interacting with a conserved water in the extended hinge region. Finally, the C-7 tertiary carbinol was waterexposed in the extended hinge region as predicted, stabilizing a water network that also included the quinazolinone carbonyl. Although several analogs displayed in Table 2 showed promising plasma concentrations in mouse, none displayed a superior profile for progression compared to the parent quinazolinone 6. With the low aqueous solubility (10 0.014b 0.55 ± 0.1 2.06 ± 0.24

IC50 values are shown as mean ± standard deviation values of at least three determinations. bIC50 values are shown as single determinations.

Table 6. Partial List of in Vitro Profiling Data for Compound 6a parameter protein bindinga mutagenicity hERG (patch clamp) CYPb inhibition (IC50)a aqueous solubility FaSSIF solubilityc FeSSIF solubilityd PAMPA (pH 5.5 and 7.4)a Caco2 permeabilitya (apical to basolateral) Caco2 permeabilitya (basolateral to apical)

result 99.4% in human serum 99% in mouse serum Ames negative 55% inhibition at 10 μM >40 μM 1A2, 2B6, 2D6, 3A4; >8 μM 2C9, 2C8

Small Molecule Reversible Inhibitors of Bruton's Tyrosine Kinase (BTK

Recommend Documents